Search outcomes
The database search yielded 585 papers and an additional eight papers were retrieved from the reference list search. Removed were 344 papers, which were duplicates, leaving 249 papers. Paper titles and abstracts were scrutinised for relevance by two reviewers (TAM, CM) and 49 papers were retained. The full text of these 49 papers was examined and checked against the inclusion criteria by the same two reviewers. Thirty three papers were excluded for illegibility. Three more papers that met the inclusion criteria for the second search were included, whereupon a total of 19 papers were selected. The updated search revealed three more studies, culminating in the inclusion of 19 papers as part of this review (see Fig. 1).
Study characteristics
This review presented 19 papers, which included 20 studies, as one paper reported on two different studies [41]. These studies, from 13 nations (Arab [42, 43], Hong Kong [44], Iran [7, 45], Israel [13], Japan [46, 47], Korea [48], Mexico [49, 50], Norway [51], Sir Lanka [19, 52], Spain [53], Sweden [54], Turkey [41, 55] and Serbia [56]), met the inclusion criteria for this review.
All studies used observational design: cross sectional (n = 5), case control (n = 9) and cohorts (n = 6). The last 6 years (2012–2017), have seen an increased interest in autism screening, as 16 of the studies included in this review were from that period, compared with four studies from between 2004 and 2011. The study sample size varied from 100 [45] to 12,984 [51], with a mean of 2207. This sample included both sexes, aged from 1 to 13 years, with a mean age of 2.6 years. The majority of studies (n = 15) were conducted in clinical settings (e.g., primary, psychiatry and hospital) [19, 43, 45,46,47, 51, 53,54,55,56]. Two studies were done in the community (day care, kindergarten, preschool centres and public primary schools) [7, 13], while three studies consisted of a combination of settings.
Almost 80% of the studies used the Modified Checklist for Autism in Toddlers (M-CHAT n = 15), as well as its revised version, with the follow-up interview (M-CHAT R/F; n = 1) as a screening instrument. However, other screening instruments, such as the First Year Inventory (FYI) in Israel [13], Social Responsiveness Scale-Preschool (SRS) in Mexico [49], Quantitative Checklist for Autism in Toddlers (Q-CHAT) in Iran [45], and Pictorial Autism Assessment Schedule (PAAS) were also recognised in this review.
Parents were the main informants in all studies, especially mothers, although in some cases (n = 6) a trained assessor, such as a medical/health science student, nurse, family physician or psychologist was also involved. The trained assessor involvement was for validation purposes, or to meet cultural preferences [44, 47, 48, 53,54,55, 57, 58]. Training (seminars, workshops and a special study module) or aids (pamphlet slides, oral presentations, instruction booklets) were used to promote awareness of autism among both professionals and parents. However, the nature of implementation, training programmes, assessors’ roles, and detailing such awareness varied from author to author and was not fully documented.
Cultural adaptation
The EVF model [33] was used to investigate the extent of cultural adaptation within the study. This model suggested addressing eight dimensions when culturally adapting an intervention. They are explained in following sections.
Language
The first dimension was that of language; placing particular attention on presenting clear and understandable language, idioms, regional words and slang, in both written and verbal forms. In this review, all studies undertook language adaptation. Each study attempted to present culturally appropriate language (verbal and written) as part of their adopted instrument, as well as in the follow-up interview. Despite similarities in linguistic adaptation procedures, studies varied in the way findings were reported. Only two studies detailed, in full, the steps involved in linguistic adaptation, such as translation, back translation, number of translators, piloting and committee review. Discussions were supported with examples [48, 53]. Seventeen studies reported some of the previous steps, most commonly, the back and forth translation [7, 13, 19, 41,42,43,44,45,46,47, 49,50,51,52, 54,55,56]. However, the translation procedure and cultural adaptation guidelines, if indeed any such guidelines were adopted, were not discernible. The exception being Nygren et al. [54] who highlighted information regarding the use of recommended guidelines for translation. In addition to translation, Seif Eldin et al. [42] incorporated different dialects from nine Arabic countries. This information was integrated into the adapted version of the Modified Checklist for Autism in Toddlers (M-CHAT), in order to promote parental understanding of autism in those countries. Perera et al. [19] attempted to conceptualise screening items in their original language, then combined each item with a photograph to facilitate parental comprehension. This step was followed by a clarity check from a random sample of professionals and members of the public.
Metaphors
This dimension addressed incorporating verbal (e.g., folk saying) and visual symbols such as images, pictures, or figures in the screening process to convey a meaning tailored to the cultural values. In this review, two studies used culturally relevant metaphors within the screening process. For example, Canal-Bedia et al. [53] developed a Spanish version of the M-CHAT and, after piloting, included an adaptation, using Spanish cultural idiosyncrasies. Items 3, 5 and 23 were modified to include examples of Spanish toys. Perera et al. [19] introduced photographs within their screening instruments, to illustrate the text of the screening items and to promote understanding.
Person
This dimension addresses the ethnic or interactional match between study participants and the investigator. “Person” was only considered on two occasions. The first was in Ben-Sasson and Carter [13], where only parents who were proficient in Hebrew were involved, which enabled them to complete the adapted version of FYI in the Hebrew language and culture with more ease. Perera et al. [19] used local children’s photographs to promote parental understanding of the screened items.
Contents
In some studies, the culture, values, customs and traditions of participants were integrated into the content of the adapted instruments and/or screening process. Only one study investigated the value of cultural information from the study groups and incorporated that into the screening instrument. Specifically, Wong et al. [44] modified the original instrument, from a checklist format to a graded score system, as a result of a pilot study which found many Chinese parents struggled to answer the original yes/no questions. The modified CHAT-23 involved selecting answers, such as “never”, “seldom”, “usually,” and “often”.
Concepts
Ten studies described how the authors’ framed the adapted instruments into formats that were more understandable and consistent with the specific culture and context. This involved re-wording, describing and generalising/specifying difficult concepts or supporting them with clarifying examples, in a written or demonstrable format, or by deleting confusing, less well understood items. For example, in Canal-Bedia et al. [53] three items from the screening instrument were re-worded after piloting to promote parental understanding (items 5, 8, and 17). Albores-Gallo et al. [50] described the meaning of the “peek-a-boo” game because some parents, such as Mexican parents, did not have a name for it. In Kamio et al. [47] and Kondolot et al. [55], trained interviewers provided parents with specific examples for each failed item, to facilitate a better understanding and enable them to judge their responses. Samadi and McConkey [7] provided a general definition for some items, when translated to the Kurdish and Persian languages, to promote parental understanding. For example, item 9, “finger flicking,” was presented in the Kurdish instrument as “any unusual finger and hand movements.” Item 10, ‘fearful behaviours’, was explained during the follow-up interview, as reactions to social situations and new experiences. Seung et al. [48] also re-worded three items (3, 5 and 11), and included examples for each and deleted the three most confusing and misunderstood items (4, 8 and 22). More explicit words for a number of unspecified items were also included, to promote instrument adequacy and an understanding for Korean parents, consistent with Nygren et al. [39], who used interpreters to describe items 11, 22, and 23. Perera et al. [19] incorporated relevant photographs within their study, to clarify item concepts and improve recognition.
Goals
From the studies reported here, it was not possible to identify whether the screening goals were constructed within the context and knowledge of values, customs and traditions, or if there was any similarity among the assessors and participants in terms of screening being desirable within the study context. This is with the exception of one study in Spain, in which the authors reported that both professionals and parents expressed an interest in routine autism screening [53].
Methods
Methods takes into consideration the incorporation of cultural knowledge into screening methodology. Five studies incorporated cultural knowledge and modified screening methods to ensure screening falls within that country’s cultural context. For example, a study by Kara et al. [41], found that when Turkish parents filled in the M-CHAT [59], 49% of participants’ screened were positive for autism. As a result, in the second study trained nurses and psychologists interviewed parents to completed the Turkish version of the M-CHAT questionnaire, where they were able to probe and clarify issues. This method proved more effective and followed a recent study [55] where the M-CHAT [59] was completed using information gathered in face-to-face interviews. This, again, was found to be useful in the Turkish culture and resulted in fewer false-positive screening results. Another example of methodological modification to meet cultural preferences and improve instrument reliability, was in a study by Wong et al. [44], where Chinese parents did not complete the entire questionnaire checklist. An observational section, completed by a trained assessor, was found to reduce false-positive results. For the same purpose, other studies incorporated the screening instrument, M-CHAT, with different instruments [52, 54], or with a follow-up interview, to enhance reliability and meet cultural needs.
Context
Context is the last dimension of the framework and takes into account the contextual issues that may affect the screening process within each culture. This review found authors of the described studies attempted to address issues which might have challenged autism screening and they suggest potential efforts to overcome these challenges. For example Kara et al. [41], Kondolot et al. [55], identified a context issue among the Turkish population: the general population was not used to completing checklists and, hence, preferred verbal interview formats. Low and middle-income families in Turkey may also have difficulty in understanding the written questionnaire. The number of years spent in education is lower (not specified) in Turkey than in Western nations. Seif Eldin et al. [42] produced an Arabic version of the M-CHAT to screen children for autism in nine Arabic countries. Participating countries were classified into four sub-groups (the Gulf area, East Mediterranean, Egypt and Tunisia) based on cultural, ethnic, political and social structure similarities, in order to reduce the impact of cultural diversity and help generate concrete conclusions. However, the authors did not report how they accounted for other cultural influences.
Feasibility
In Bowen et al. [39] taxonomy of feasibility constructs were used to evaluate feasibility aspects for each study. Some information relevant to these aspects was identified and detailed in the following sections for each dimension.
Acceptability
With the exception of one study, the perception of suitability or satisfaction towards autism screening was not documented. In their two-phase study, Canal-Bedia et al. [53] adapted and validated the M-CHAT for the Spanish population, highlighted the “great interest” that both parents and professionals showed in routine screening for communicative and social development in Spain.
Demand
Only one study documented interest in using autism screening within their current practice. Nygren et al. [54] trained doctors and nurses in child health care settings to screen children for autism, within the two and a half years of age check-up window. The study highlighted that the trained assessors continued to use their newly acquired skills to refer suspected cases of autism (in children both younger and older than two and a half years) for evaluation, even after completion of the study.
Implementation
Although the studies varied in design, purpose and results, screening for autism seemed to be successfully implemented, as planned, for the intended participants. However, the studies investigated here varied in the detail of the implementation process. Five studies provided full detail of the planning and implementation process associated with screening [13, 44, 48, 51, 53]. The remaining studies briefly explained what they had undertaken [7, 19, 42, 43, 45, 46, 49, 50, 52, 55, 57]. The shortened explanations might be the result of journal word limits.
Practicality
Most studies reported that screening instruments identified autism, but expressed concern over their adequacy in population-based settings. Studies also highlighted the cost burden of vetting instruments [13, 43, 46, 49,50,51, 53] and the interventions required to redress limitations, like training assessors and employing follow-up interviews [50, 55]. Among all screening instruments, the M-CHAT and revised versions, including follow-up interviews, were adopted by almost 80% (n = 16) in the studies reviewed. M-CHAT was implemented either separately or with another instrument (Checklist for Autism in Toddlers (CHAT), Early Screening of Autistic Traits Questionnaire (ESAT), Joint Attention Observation (JA-OBS), CBCL/15.5-5 Hiva and/or a follow-up interviews) for cultural preferences or validation purposes. Despite the disparity in implementation, analysis and adaptation methods, similarities were noted in the practical features of M-CHAT across numerous studies (i.e., time and key identifers). For example, studies reported that the M-CHAT can be completed either by a parent or by an assessor within 5–10 min and the follow-up interview would need a further 10 min. Interestingly item 13 “imitate you” was found to be the only key identifier item from the original M-CHAT (i.e., can discriminate between children with or without ASD) across nations, with some variation in strength for the identification of autism. The reviewed studies also presented the differences in other key identifer items from the original M-CHAT, such as item 21 “understanding” [41, 46, 47, 50, 53, 54], and item 23, “checking reaction” [44, 46, 47, 53, 54]; while item 11, “over-responsiveness to noise” presented some concerns in five studies [41, 43, 48, 53, 54].
Besides M-CHAT, this review identified other instruments that lent themselves to being completed by parents in a short time frame. For example, the First-Year Inventory (FYI) includes 60 items, takes about 20 min to rate the 60 items as: never, seldom, sometimes and often, or includes multiple choice questions to identify children at risk of autism or a related developmental disability. Similarly, SRS, a 65-item rating scale, ranging from 1 (not true) to 4 (almost always true), requires 15–20 min to complete. In contrast, Quantitative Checklist for Autism in Toddlers (Q-CHAT), scored on a 5-point scale (0—never to 4—always) contains 25 items, and takes 5–10 min to complete. Finally, the 21 PAAS items with “yes” or “no” choices, can be completed in 15–20 min.
Adaptation
Adaptations were made in all studies, with variations to accommodate cultural values and traditions, depending on the study aims and perspective.
Integration
Integrating the screening process into an existing system was common among studies but is not encouraged. The studies suggest the possibility of introducing autism screening at the primary level (paediatric, surveillance programme and routine practice) [13, 42, 44, 46, 47, 50, 52,53,54, 56], psychiatric level [50] or within a school setting [45]. However, they also warned of potential instrument inadequacy, as well as any cultural or demographic influences on the screening context. Some studies also noted the importance of recognising individual health system needs and capacities, prior to introducing mandatory screening programmes [44, 52, 53, 55, 56].
Expansion and limited efficacy
Most studies did not encourage autism screening beyond the study context and indicated limited efficacy in adapting the instruments for different populations. Results of studies varied, making it very difficult to compare them internationally and formulate conclusions. For example, studies adapted various screening instruments (M-CHAT, M-CHAT R/F, Q-CHAT, CHAT-23, SRS, FYI and PAAS) that represented diverse levels of psychometric properties (i.e., reliability and validity of the instrument) [13, 19, 44, 49, 51, 54]. Even within studies that used the same instrument (M-CHAT), variations of responses to the instrument items, key identifiers [42, 44, 48, 50, 54], and instrument adequacy were reported [13, 46, 50, 51].